Characterization of cephalosporin-resistant Escherichia coli from Norwegian broiler production

Masteroppgave 2018 60 stp

Institute for chemistry, biotechnology and food science Faculty of veterinary medicine and bioscience

Knut Rudi

Characterization of cephalosporin- resistant Escherichia coli from Norwegian broiler production.

Kingsley Osei Oteng

Master’s in Biotechology Microbiology

Preface

The work presented in this Master thesis was performed under the Section for food Safety and Emerging Health Threats at the Norwegian Veterinary Institute (NVI) in Oslo. This study was part of the NoResist project, which main objective is to “to combat antimicrobial resistance in the Norwegian food chain”.

First, I would like to thank my supervisor Solveig Sølverød Mo for giving me the opportunity to work on this project and giving me some guidance during the laboratory work.

Second, I would like to thank Marianne Sunde, another co-supervisor for giving me some insightful advices regarding my thesis. Moreover, I would like to show my appreciation to Camilla Sekse and Jannice Schau Slettemeås for assisting me with the bioinformatics analysis. Another thanks to all the laboratory engineers at the Microbiology section at NVI, especially Bjørg Kvitle for sharing your valuable time with me and knowledge. Finally, I would like to show a great appreciation to my mum for her prayers and support. Finally, a big thanks to Clay Gouin at the NMBU writing center for giving me some guidance on my writing and my roommate Bastien Pierre Bissaro for all the help.

Abstract

Recently, extended spectrum beta lactamase (ESBL)-producing E. coli with a new genotype (blaCTX-M-1) emerged in the Norwegian broiler production pyramid. Occurrence of ESBL-producing bacteria such as E. coli is a global concern because they resist highly important antimicrobials known as third-generation cephalosporins used in human medicine. Development of ESBL-

producing bacteria have resulted from the use and misuse of antimicrobials in humans and livestock production. However, Norway hardly use antimicrobials in its broiler production but still detected ESBL-producing E. coli with blaCTX-M-1. Thus, the goal of this study was to examine three questions regarding ESBL-producing E. coli emergence in the Norwegian broiler production pyramid. First, was this emergence a clonal spread of a specific ESBL-E. coli variant with one plasmid? Second, could it have been the clonal spread of several genetically unrelated E. coli variants with different plasmids and third, did a horizontal transfer of blaCTX-M-1-harboring plasmids occur between different E. coli STs? Together 35 ESBL-producing E. coli isolates were investigated.

To examine these questions, molecular typing methods including PCR-based phylotyping and pulsed-field gel electrophoresis were used to analyze genetic relatedness between the ESBL- producing E. coli isolates. Isolates’ phenotypic resistance was determined by minimum inhibitory concentration -and disk-diffusion tests. Plasmids associated with blaCTX-M-1 were identified through conjugation experiments and plasmid replicon typing. For further characterization of the isolates, whole genome sequencing was performed to determine multi-locus sequence types (MLSTs), serotypes, virulence genes, acquired antimicrobial resistance genes, and genetic relatedness based on single nucleotide polymorphisms. Using the whole genome data, blaCTX-M-1-carrying plasmids were identified, and one blaCTX-M-1-IncI1α plasmid re-constructed.

Results showed most of the E. coli isolates were genetically related and grouped into a large cluster represented by the phylogroup D/ST-57-O140:H25 clonal lineage. Moreover, E. coli isolates from parents seemed more genetically diverse than the broiler isolates. IncI1α grouped into plasmid multi-locus sequence types (pMLST) 3 and 7 were identified as the main blaCTX-M-1-carrying

plasmids. IncI1α plasmids from this study shared close homology with other IncI1α plasmids detected in broilers from France and Switzerland. Genetic characteristics of the isolates and plasmids identified in this study were similar to previous reports in broilers from several European countries. Thus, the results demonstrated both clonal dissemination and horizontal transfer of the IncI1α plasmids disseminated cephalosporin resistant E. coli in the Norwegian broiler production.

The blaCTX-M-1-IncI1α plasmid characterized carried a toxin component, hok gene that could have maintained IncI1α plasmid in E. coli in the broiler production pyramid.

Norwegian Abstract

Nylig ble det gjort funn av ESBL-produserende E. coli med en ny genotype (blaCTX-M-1)i den norske slaktekyllingproduksjonen. Forekomst av ESBL-produserende bakterier som E. coli er en global bekymring fordi de er motstandsdyktige mot kritisk viktige antimikrobielle midler som tredje generasjons-cefalosporiner, som brukes til behandling av infeksjoner hos mennesker.

Utviklingen av ESBL-produserende E. coli skyldes bruk og misbruk av antimikrobielle midler hos mennesker og dyr. Imidlertid bruker Norge knapt antimikrobielle i sin slaktekyllingproduksjon, men allikevel påvises ESBL-produserende E. coli med blaCTX-M-1. Hensikten med denne oppgaven var derfor å undersøke tre problemstillinger angående forekomsten av ESBL-produserende E. coli i den norske slaktekyllingproduksjonen. Først, kunne forekomsten ha vært en klonal spredning av en ESBL-produserende E. coli variant med ett plasmid? For det andre, kunne det ha vært en klonal spredning av flere genetiske ulike E. coli varianter med ulike plasmider, og sist, skjedde det en horisontal overføring av blaCTX-M-1-bærende plasmider mellom forskjellige E. coli sekvenstyper? Til sammen ble 35 ESBL-E. coli med blaCTX-M-1 undersøkt.

For å undersøke disse problemstillingene, ble det anvendt molekylære metoder som PCR- basert fylotyping og puls-felt gelelektroforese for å fastslå det genetiske slektskapet mellom de ESBL-produserende E. coli isolatene. Isolatenes fenotypiske resistens ble bestemt ved minste hemmende konsentrasjon og lappediffusjonstest. Plasmider assosiert med blaCTX-M-1 ble identifisert gjennom konjugasjonsforsøk og plasmid-replikon-typing. For videre karakterisering av isolatene, ble de helgenomsekvensert for å bestemme MLST, serotype, virulensgener, antibiotika

resistensgener og genetisk slektskap basert på enkletnukleotidpolymorfier (SNP). Ved bruk av helgenomdata ble blaCTX-M-1-bærende plasmider identifisert, og ett blaCTX-M-1-IncIl-plasmid karakterisert.

Resultatene viste at de fleste av E. coli-isolatene var genetiske relaterte og gruppert i et stort kluster representert av fylogruppe D/ST-57-O140: H25 klonen. I tillegg så det ut som at det var en større genetisk variasjon i E. coli-isolatene fra foreldredyr enn isolatene fra slaktekylling.

IncI1α/ST3 og IncI1α/ST7 ble identifisert som de viktigste blaCTX-M-1-bærende plasmidene. IncI1α plasmider fra denne studien var i større grad likt andre IncI1α plasmider funnet i slaktekylling fra Frankrike og Sveits. De genetiske egenskapene til isolatene og plasmidene funnet i denne studien lignet på isolater og plasmider fra slaktekylling i flere europeiske land. Stort sett indikerte

resultatene at både klonal utbredelse og horisontal overføring av IncI1α plasmidene spredte

cefalosporin resistent E. coli i den norske slaktekyllingproduksjonen. Det karakteriserte blaCTX-M-1- IncI1α plasmidet uttrykte en toksinkomponent, hok-genet, som kunne ha opprettholdt IncIl-plasmid i E. coli i slaktekyllingproduksjonen.

Table of contents

Preface ... I Abstract ... II Norwegian Abstract ... III Table of contents ... III

Introduction ... 1

1. ... 1

1.1 Norwegian broiler production ... 1

1.2 Antimicrobial agents: ... 1

1.2.1 Inhibition of cell wall synthesis: ... 2

1.2.2 Inhibition of protein synthesis ... 3

1.2.3 Inhibition of nucleic acid synthesis ... 3

1.2.4 Inhibition of metabolic processes ... 3

1.3 Antimicrobial resistance ... 4

1.3.1 Main resistance mechanisms ... 4

1.3.1. i Enzymatic inactivation of antimicrobials ... 4

1.3.1. ii Efflux pumps ... 5

1.3.1.iii Alteration in target sites ... 5

1.4 Acquired resistance via horizontal gene transfer ... 6

1.4.1 Transformation ... 6

1.4.2 Transduction ... 6

1.4.3 Conjugation ... 6

1.5 Mobile genetic elements (MGEs): ... 7

1.5.1 Plasmids ... 7

1.5.2 Transposons... 7

1.5.3 Integrons and gene cassettes ... 8

1.5.4 Insertions sequences ... 8

1.6 Beta-lactamases ... 9

1.6.1 Extended Spectrum Beta-Lactamase (ESBLs) ... 9

1.6.2 AmpC beta-lactamases: chromosomal AmpC (cAmpC) and plasmid- mediated AmpC (pAmpC) ... 10

1.7 Characterization of cephalosporin-resistant E. coli isolates ... 10

1.7.1 Phylogenetic grouping of E. coli isolates ... 10

1.7.2 Molecular typing of E. coli isolates: PFGE ... 10

1.7.3 Plasmid characterization ... 11

1.8 Epidemiology of cephalosporin-resistant E. coli and CTX-M genes ... 11

1.8.1 Epidemiology in Norwegian broiler production pyramid and Europe ... 12

1.9 Public health: ramifications of ESBL-producing E. coli ... 14

1.10 Aim of study ... 15

2 Materials and methods. ... 15

2.1 Selective isolation and confirmation of ESBL-producing E. coli ... 15

2.2 Characterization of ESBL-producing E. coli isolates ... 16

2.2.1 DNA extraction ... 16

2.2.2 Phylogenetic grouping ... 16

2.3 Antimicrobial susceptibility testing ... 18

2.3.1 Micro broth dilution for MIC determination ... 18

2.3.2 Agar disk diffusion method ... 19

2.4 Pulsed-Field Gel Electrophoresis (PFGE) ... 19

2.4.1 Preparation of cell suspension ... 19

2.4.2 Gel plug preparation ... 19

2.4.3 Lysis and washing of plugs ... 20

2.4.4 Digestion of DNA in agarose plugs ... 20

2.4.5 Gel electrophoresis ... 20

2.4.6 Staining and visualization of the gel ... 20

2.4.7 Data analysis of the gel image ... 20

2.5 Conjugative transfer and characterization of plasmid replicons ... 21

2.5.1 Conjugation experiments ... 21

2.5.2 PCR-detection of blaCTX-M-1 in the transconjugants ... 21

2.5.3 Plasmid typing: PCR-based replicon typing (PBRT) of transconjugants .... 22

2.5.4 Suceptibility testing of transconjugants ... 23

2.6 Whole genome sequencing (WGS) ... 23

2.6.1 Bacterial DNA isolation with Qiasymphony ... 23

2.6.3 Pre-processing of raw sequence data: assembly of genomes ... 24

2.7 In silico analysis of the whole genome sequencing: ... 24

2.7.1 Bacterial Analysis Pipeline: ... 24

2.7.2 Phylogenetic analyses: CSI-phylogeny ... 26

2.7.3 Serotyping (SerotypeFinder) ... 26

2.7.4 blaCTX-M-1-harboringplasmid characterization ... 26

2.7.4 i DNA extraction with QiaAMP DNA kit... 27

2.7.4 ii PCR and Sanger sequencing ... 27

2.7.5 Annotation of the plasmid ... 27

3. Results ... 28

3.1 Genotypic and phenotypic characterization of the E. coli isolates ... 28

3.1.2 Antimicrobial resistance profile: phenotypic testing ... 29

3.1.2.i MIC determination ... 29

3.1.2.ii Agar Disc diffusion test ... 30

3.1.3 PFGE ... 31

3.2 Characterization of blaCTX-M-1-carrying plasmids ... 33

3.2.1 Conjugative transfer of blaCTX-M-1 and plasmid replicon typing ... 33

3.2.2 PBRT ... 33

3.2.3 Susceptibility testing of transconjugants ... 35

3.3 Whole genome data analysis: in silico characterization of the E. coli isolates... 35

3.3.1 Multi-locus sequence typing (MLST) ... 35

3.3.2 Phylogenetic analysis of the E. coli isolates ... 36

3.3.3 Detection of antimicrobial resistance genes ... 37

3.3.4 Detection of virulence genes ... 38

3.6.5 Serotyping of the E. coli isolates ... 40

3.6.6 Whole genome plasmid typing ... 41

3.6.7 pMLST of IncI1α plasmids ... 42

3.6.8 IncI1α plasmid characterization ... 43

3.6.9 Isolates with non-transferrable plasmids ... 46

3.6.10 MDR isolates ... 46

4. Discussion ... 47

4.1 Phylotyping, PFGE, and MLST, serotyping of isolates, SNP-analysis ... 47

4.2 Plasmid typing: conjugation, PBRT, PlasmidFinder and pMLST. ... 49

4.3 Phenotypic resistance testing and acquired resistance genes (ResFinder): ... 50

4.4 Virulence genes associated with the different E. coli MLSTs/phylogenetic groups: ... 51

4.5 Isolates with non-transferrable plasmids ... 51

4.6 Multi-resistance isolates ... 51

4.7 Non-typeable isolates PFGE ... 52

4.8 Limitations of study ... 52

5 Conclusion ... 53

6. References ... 53

7. Supplementary ... 58

1. Introduction

1.1 Norwegian broiler production

Broiler (Gallus gallus) is a chicken raised specifically for meat production. In Europe, the broiler production follows a breeding pyramid structure where purebred animals such as great grandparents are on top followed by the breeders (grandparent and parent stocks) in the middle and the broilers at the bottom of the pyramid (Figure. 1) (Mo, 2016). In Norway, the parent animals and broilers are raised, with the number of broilers produced in 2017 around 65.5 million according to the Statistics Norway (SSB, 2017)

Broiler production in Norway starts with the import of hatching eggs from grandparent animals, exported from Scotland to Sweden. The imported eggs from Sweden are hatched in Norway into day-old parent animals and sent to rearing farms where they are kept until 18 weeks old. Eggs from the parent animals that are 18 weeks or older are hatched into day-old broilers at a hatchery and further delivered to the broiler farms. Here, broilers are raised until 28-32 days before being slaughtered for meat production (Mo, 2016).

Figure 1. The pyramidal structure of broiler production. Adapted from (Mo, 2016).

1.2 Antimicrobial agents:

Antimicrobial agents used in animal production to prevent risk of infections and treat diseases are the same drugs classes used in human medicine. These drugs function by

targeting different structures in the bacteria andare classified based on their ability to inhibit

cell growth (bacteriostatic) or induce cell death (bactericidal) (Kohanski et al., 2010). To achieve the bacteriostatic or bactericidal effects, antimicrobial agents interfere with the reactions that synthesize structures that bacteria depend on to survive and develop. Thus, the cell wall, ribosomes, and nucleic acids are the main targets of these antimicrobial agents, particularly the bactericidal. On the other hand, the bacteriostatic drugs prevent bacteria from carrying out their metabolism (Kohanski et al., 2010).

Figure 2. Schematic representation of bactericidal antimicrobial agents and its mechanism leading to cell death. Adapted from (Kohanski et al., 2010).

1.2.1 Inhibition of cell wall synthesis:

The cell wall is the most vital structure that surrounds bacteria. It strengthens and protects the cell against stress and damage. Without this structure, bacteria are more susceptible to attack by various toxic compounds such as antimicrobials. The cell wall is the primary target of the beta-lactam antimicrobials. In the presence of beta-lactams, bacteria are unable to synthesize new cell wall as the enzymes (transpeptidases) involved are repressed (Kohanski et

al., 2010). As a result, the cell wall structure weakens leading to lysis and cell death (Figure 2).

1.2.2 Inhibition of protein synthesis

The site of protein synthesis in bacteria, ribosomes, are the main target of antimicrobial agents such aminoglycosides and chloramphenicol. Ribosomes consist of two subunits: 50S and 30S that are involved in protein synthesis (Figure 2). Once antimicrobial agents

compromise ribosomes, reactions involved in the protein production stops or the bacteria generate non-functional proteins. For instance, chloramphenicol prevents peptide bond formation that result in a functional protein. On the other hand, aminoglycoside interacts with the 30S ribosome subunit and causes the tRNA to carry the incorrect amino acids to the ribosomes (Kohanski et al., 2010, Willey et al., 2014b)

1.2.3 Inhibition of nucleic acid synthesis

Synthesizing nucleic acids (DNA) is a fundamental process in all life forms, including bacteria. To initiate this process in bacteria, it requires a “relaxed” DNA where the double strands are broken, and twists removed. The DNA gyrase (topoisomerase II) and

topoisomerase IV are the main enzymes that bind to relax the DNA (Willey et al., 2014b).

Thus, antimicrobial agents called quinolones target these enzymes to inhibit their function and, in the process, DNA synthesis.

Quinolones restrict the DNA gyrase and topoisomerase IV by forming a stable complex structure with them (Figure 2). In this structure, the enzymes become trapped and are unable to relax the condensed DNA. As a result, the DNA synthesis is blocked and cell growth prevented. Furthermore, quinolones are bactericidal and cell death occurs when the drugs inhibit a DNA repair system known as the SOS response in bacteria (Figure 2).

1.2.4 Inhibition of metabolic processes

Some antimicrobial agents are known as “anti-metabolites” because they interfere with metabolic pathways that are essential to the bacteria (Scholar and William, 2000). Folic acid synthesis is essential in bacteria because the process generate folic acid, which bacteria use to synthesize DNA, RNA, and other cell components such as ATP. Sulfonamides and

trimethoprim are well-known anti-metabolites that disrupt the folic acid synthesis in bacteria (Willey et al., 2014b).

1.3 Antimicrobial resistance

According to the World Health Organization (WHO), antimicrobial resistance (AMR) remains one of the major threats to global health (WHO, 2014). Antimicrobial resistance is a natural phenomenon in microorganisms, such as bacteria, where antimicrobials designed to treat diseases in humans and animals become futile. Over the years, antimicrobial resistance has spread worldwide, and one major cause involves the inappropriate use of antimicrobials in human and veterinary medicine (Ventola, 2015). As a result, these actions have driven the emergence and spread of antibiotic resistant bacteria destroying important antimicrobials (ECDC & EFSA, 2018). Of the antimicrobials, resistance to third-generation cephalosporins greatly concerns the WHO as they prioritize these drugs as “critically important” in human and animal therapy (WHO, 2017).

In general, antimicrobial resistance in bacteria develops by two main mechanisms- intrinsic and acquired resistance (Smith and Lewin, 1993). Intrinsic resistance occurs when bacteria naturally resist antimicrobials due to its structure or functional processes (Blair et al., 2015). For instance, the Mycoplasma spp. are “intrinsically resistant” to beta-lactams because they target the cell wall, which these bacteria do not possess (Thenmozhi et al., 2014b).

Another example of intrinsic resistance involves the vancomycin resistance in Gram-negative bacteria (Thenmozhi et al., 2014b). The outer membrane structure in these bacteria acts as a barrier against the drug entry and prevent it from reaching its target site (Cox and Wright, 2013). In contrast, acquired resistance occurs when a susceptible bacterium becomes resistant to an antimicrobial (WHO, 2011).

Of the two resistance mechanisms, acquired resistance is clinically relevant because it can spread among different bacteria species and reduce antimicrobial treatment options (Munita and Arias, 2016) . Furthermore, acquired resistance develops in two ways, i.e., when there is a genetic mutation associated with the antimicrobial’ actions or when a bacterium obtains foreign DNA consisting of resistance genes from other bacteria (Munita and Arias, 2016)

1.3.1 Main resistance mechanisms

1.3.1. i Enzymatic inactivation of antimicrobials

Bacteria can produce enzymes that either destroy or modify the antimicrobials. One example of antimicrobial inactivation due to enzymes is the production of beta-lactamases that destroy beta-lactams. One the other hand, bacteria can generate modifying enzymes (acetylases, phosphorylases, and adenylase) that cause a steric hindrance in the antimicrobial

molecule. The steric hindrance effect reduces the drug’s affinity for its the target sites (Munita and Arias, 2016).

1.3.1. ii Efflux pumps

Antimicrobials unable to enter the cell to perform its actions can be due to efflux pump production in bacteria (Cox and Wright, 2013). Efflux pumps are proteins that transport antimicrobials or other toxic compounds out of the cell (Figure 3). Genes encoding efflux proteins are located on chromosomes or mobile genetic elements (MGEs) (Thenmozhi et al., 2014a). Tetracycline resistance in certain bacteria illustrates an efflux-mediated because the tetracycline efflux protein (TetA) pumps out the drug before reaching its target in the cell (Munita and Arias, 2016).

Figure 3. Representation of an efflux-mediated resistance. Both antimicrobials A and B enter the cell membrane, but antimicrobials B is transported out from the cell by efflux pump Adapted from (Blair, 2014).

1.3.1.iii Alteration in target sites

For antimicrobials to exert its functions, they need to interact with their target sites in the bacteria. However, bacteria can prevent this interaction by protecting or modifying these target sites (Munita and Arias, 2016). One way a bacterium modifies an antimicrobial target site is to mutate the genes encoding that site. In turn, the genes will encode abnormal targets sites that cannot interact with the antimicrobials. Rifamycin resistance due to point mutations in the rpoB gene shows an example of mutational resistance (Munita and Arias, 2016).

1.4 Acquired resistance via horizontal gene transfer

In addition to genetic mutation, exchange of genetic material between bacteria is the process where they can acquire an antimicrobial resistance. This process is generally referred to as horizontal gene transfer and occurs via three main mechanisms: transformation,

transduction, and conjugation (Pepper et al., 2014).

1.4.1 Transformation

In the transformation mechanism, bacteria take up DNA from their external environment and integrate into its genome via homologous recombination (Pepper et al., 2014). The DNA is released by dead cells referred to as donors. Bacteria that can take up the DNA and

transform into the donor cell are referred to as competent bacteria.

1.4.2 Transduction

Transduction, on the other hand, rely on bacteriophages to transfer DNA between bacteria.

Bacteriophages are viruses that infect bacteria and further use them as host to multiply and produce more bacteriophages. After viruses multiply, they assemble into mature virions and at this stage, they can take up DNA fragments containing antimicrobial resistance genes from the bacteria (Willey et al., 2014a). When these bacteriophages infect new bacterial host cells, they inject piece of the DNA from the previous bacterial host cell into the genome of a new bacteria host (Pepper et al., 2014).

1.4.3 Conjugation

Of the three mechanisms, conjugation is most effective in transferring antimicrobial resistance between bacteria. This mechanism requires a direct cell-to-cell contact between two cells, i.e. a donor and recipient (Figure 4). During conjugation, the donor cell (F⁺) that

harbours the resistance genes on their mobile genetic elements (MGEs) or conjugative elements, transfer them to the recipient cell (F^-). The gene transfer from the donor cell to recipient is mediated by transfer (tra) genes that form sex pili to connect the two cells. After receiving the MGEs containing the resistance genes, the recipient cell becomes a

transconjugant that in turn can transfer the resistance genes to other recipient cells (Wiley et al., 2014). Bacteria such as E. coli carry out conjugation to transfer genes to other bacteria, as seen with case of cephalosporin resistance (Mo et al., 2017).

Figure 4. Illustration of a conjugation process. DNA is transferred from donor cell (F⁺, left) to recipient cell (F^-, right) through the sex pilus. Adapted from (Willey et al., 2014a).

1.5 Mobile genetic elements (MGEs):

Mobile genetic elements, i.e. plasmids, transposons, and integrons are DNA segments in bacteria that move from parts of a genome to another or between genomes (Bennett, 2008).

These elements are easily transferred between bacteria via the horizontal gene transfer mechanisms, and can harbour AMR encoding genes (Cantón et al., 2012).

1.5.1 Plasmids

Plasmids, seen as circular and double-stranded in bacteria, are additional DNA that can replicate independently from the chromosomal DNA. Furthermore, some plasmids are referred to as conjugative plasmids because they contain genes necessary for conjugation functions, i.e. the formation of sex pilus to enable horizontal gene transfer. Some plasmids have developed mechanisms, such as toxin-antitoxin (TA) systems, to ensure their

maintenance in the bacterial cell. The toxin-antitoxin system maintains plasmids in the cell by eliminating daughter cells that have lost the plasmids during cell division (Bennett, as cited in Brolund, 2014).

1.5.2 Transposons

Transposons (Tn), often known as “jumping genes”, are DNA sequences that move from one location on the genome to another. Like the conjugative plasmids, they contain genes that encode for conjugation and antimicrobial resistance. However, unlike the plasmids, they do not have their own origin of replication (oriT) region and for this reason, they

integrate themselves into chromosomes or plasmids for maintenance. Tn10 and Tn3 are known transposons that resist tetracycline and beta-lactams, respectively (Bennett, 2008).

1.5.3 Integrons and gene cassettes

Integrons are genetic elements that capture and carry antimicrobial resistance genes. They consist of three elements including (1) an integrase-encoding gene, int, that is required for site-specific recombination (2) a attI region, where DNA sequences known as gene cassettes insert, and (3) a promoter region located upstream of the attI where the gene cassettes are transcribed and expressed (Bennett, 2008). One common integron found on plasmids, chromosomes, and transposons is the class 1 integron (Figure 5).

Figure 5. Illustration of the class 1 resistance integron element. Integron 0 is considered the most basic integron without gene cassette. The class I integron contains an inserted gene cassette in addition to the four genes: sulphonamide resistance, sul1; quaternary ammonium compound resistance, qacE∆1 and the two orf 5 and 6 with unknown functions. Adapted from Adapted from (Bennett, 1999).

1.5.4 Insertions sequences

Of the mobile genetic elements, insertion sequences are the simplest due to its size (1 kb).

These elements are involved in the over-expression of bla genes and integrate within conjugative plasmids present in bacteria (Cantón et al., 2012). For instance, in the Kluyvera spp., the movement of the blaCTX-M genes has been associated with insertion sequence elements (ISEs) located upstream. Among the ISEs, the ISEcp1 is most frequent at the upstream of different blaCTX-M genes such as the blaCTX-M-1, blaCTX-M-2, blaCTX-M-9 etc. (Figure 6) (Lartigue et al. 2004).

Figure 6. Genetic environments of the different blaCTX-M genes: blaCTX-M-1, blaCTX-M-2, blaCTX-M-9, and blaCTX-M-25. ISEcp1 and other insertion sequence elements (red arrow) are present in all four genes. Adapted from (Canton et al., 2012).

1.6 Beta-lactamases

Beta-lactamases are group of enzymes produced by bacteria that degrade beta-lactam antimicrobials. These enzymes destroy beta-lactam drugs by hydrolysing the most active part, which is the beta-lactam ring. In the process, the drugs become ineffective to kill the bacteria.

According to the Ambler’s Molecular classification, beta-lactamases are grouped into four main classes: A-D (Bradford, 2001) . The most important class A beta-lactamase involves the extended spectrum beta-lactamases (ESBLs), whereas in class C, the AmpC beta-lactamases are prevalent (Bradford, 2001). Since this study is about the ESBLs, emphasis on AmpC- enzymes is limited.

1.6.1 Extended Spectrum Beta-Lactamase (ESBLs)

Extended spectrum beta-lactamases (ESBLs) are usually plasmid-mediated beta- lactamases that resist third- and fourth-generation cephalosporins, and monobactams. In contrast, they are sensitive to cefoxitin, carbapenems, and the beta-lactamase inhibitors such as clavulanic acid. ESBLs are classified into different families including the TEM-, SHV-, and CTX-M (Bradford, 2001).

During the 1990s, SHV and TEM were the most common ESBL families found in E. coli and Klebsiella spp (EFSA, 2011). However, in the early 2000s, the CTX-M families emerged, and now are the most dominant genotype found in ESBL-producing E. coli from humans, food-producing animals, food, and the environment in Europe (Seiffert et al., 2013).

1.6.2 AmpC beta-lactamases: chromosomal AmpC (cAmpC) and plasmid- mediated AmpC (pAmpC)

AmpC beta-lactamases are cephalosporinases produced by some Gram-negative bacteria that resist third-generation cephalosporins, cefoxitin and the beta-lactam inhibitors. In contrats, they are sensitive to fourth-generation cephalosporins and carbapenems (Seiffert et al., 2013). Some Enterobacteriaceae have genes encoding AmpC enzymes in their

chromosomes. Genes encoding cAmpC beta-lactamases are found on weak promoters. Here, the blaAmpC genes are expressed in low numbers and as result cannot contribute to

cephalosporin resistance (Jacoby, 2009). However, mutations in the weak promoters or attenuator regions of the chromosomal ampC gene can result in hyper-production of the cAmpC beta-lactamases and hence, resistance to third-generation cephalosporins (Shaheen et al., 2011). Over the past years, several studies have reported the movement of some AmpC- producing genes, particularly blaCMY-2, from chromosomes to plasmids, where it expresses the cephalosporin-resistance (Jacoby, 2009).

1.7 Characterization of cephalosporin-resistant E. coli isolates

In general, methods used to characterize E. coli strains possess a high discriminatory power where it distinguishes between two closely related bacterial strains (EFSA, 2011, Farber, 1996).

1.7.1 Phylogenetic grouping of E. coli isolates

Phylogenetic analysis shows E. coli is divided into four main phylogroups: A, BI, B2, and D (Picard et al., 1999). Phylogrouping of E. coli strains involves the combination of the four genetic markers: gadA, chuA, yjaA, and the TSPE.C4 DNA fragment. A multiplex PCR is the strategy used to categorize E. coli into the four main phylogroups (Doumith et al., 2012). E.

coli strains that fall into the phylogroups B2 and D are considered extra-intestinal pathogenic (ExPEC) because they possess virulence traits that cause extra-intestinal infections such as bacteremia, sepsis, urinary tract infections, and meningitis in humans (Picard et al., 1999, Smith et al., 2007). The commensal E. coli strains, on the other hand, are the phylogroups A and B1 (Picard et al., 1999).

1.7.2 Molecular typing of E. coli isolates: PFGE

Pulsed-field gel electrophoresis (PFGE) is a standard fingerprinting method used to investigate the genetic relatedness of several bacterial strains such as E. coli. Compared to

other molecular typing methods, PFGE possess a higher discriminatory power and yields more reproducible results (Sabat et al., 2013). In principle, the bacterial DNA embedded in agarose plugs is lysed and digested with a restriction enzyme, and separated on an agarose gel to generate a set of fingerprints that show the similarity or differences between the isolates (EFSA, 2011).

1.7.3 Plasmid characterization

Characterizing plasmids harboring ESBL/pAmpC genes is crucial in studying how they disseminate between different reservoirs (EFSA, 2011). Plasmids are characterized into incompatibility (Inc) groups, where incompatibility refers to the inability of two plasmids belonging to the same Inc goup to exist stably in the same bacterial cell (Thomas, 2014). The Inc groups IncF, IncI1α, IncN, IncA/C, IncK, and IncHI12 are the major plasmids that

disseminate ESBL/AmpC genes (Carloni et al., 2017). These plasmids are epidemic and have been identified in ESBL/AmpC-producing bacteria from animals, food products and humans (EFSA, 2011). The IncI1α and IncN plasmids particularly have been linked to the spread of the blaCTX-M-1 gene among E. coli from poultry (Zurfluh et al., 2014, EFSA, 2011).

1.8 Epidemiology of cephalosporin-resistant E. coli and CTX-M genes In the past years, third-generation cephalosporin resistance has increased in several bacteria (ECDC & EFSA, 2016). This has been demonstrated in E. coli strains from different reservoirs including food-producing animals, water, soils and human clinical samples (Hu et al., 2013). Production of the ESBLs/AmpC beta-lactamases by bacteria have resulted in higher rates of third-generation cephalosporin resistance (EARS-NET, 2016). Among food- producing animals, broilers are highly contaminated by ESBL/AmpC-producing E. coli and the cause of this has been linked to the massive use of antimicrobial agents in the production (WHO,2011). Further analysis shows blaCTXM-1 and blaCMY-2 as the main genes facilitating this resistance in poultry (Dierikx et al., 2013, Leverstein-van Hall et al., 2011).

On the contrary, despite low antimicrobial usage, AmpC -producing E. coli with blaCMY-2

have been detected in the Norwegian broiler production pyramid since selective screening was initiated in 2011 (NORM/NORM-VET 2011). First ESBL-producing E. coli found in broiler in Norway was reported in 2006 (NORM/NORM-VET, 2006). However, the identified ESBL-producing E. coli had a different genotype, which was blaTEM-20. Since then AmpC- producing E. coli have been prevalent until the discovery of the ESBL-producing E. coli with blaCTX-M-1. in 2016.

Worldwide, cephalosporin resistance among E. coli has been associated with the horizontal transfer of blaCTX-M-1-carrying plasmids(Cantón et al., 2012). For the past years, the study on plasmid-encoded ESBL genes has been prioritized due its spread of antimicrobial resistance between bacteria. Thus, studying the epidemiology of these plasmids in E. coli from food-producing animals can reveal if they both share common plasmid(s).

In this study, cephalosporin-resistance in E. coli from broilers in Norway and Europe are examined. In several European broiler-producing countries, ESBL/AmpC-producing E. coli have been identified (ECDC & EFSA, 2016). Countries such as France, Spain, the

Netherlands, Belgium, Italy and Switzerland have reported ESBL-producing E. coli with blaCTX-M-1 as most prevalent in broilers (EFSA, 2011), whereas in Norway, AmpC-producing E. coli with blaCMY-2 is the dominant type (Mo et al., 2017). However, in 2016, ESBL-

producing E. coli with blaCTX-M-1 was detected in broilers and parent flocks from Norway (unpublished data).

1.8.1 Epidemiology in Norwegian broiler production pyramid and Europe The first detection of an ESBL-producing E. coli with blaCTX-M-1 in poultry occurred during a Spanish antimicrobial resistance surveillance program in 2000-2001 (Briñas et al., 2003). Fast forward to today, the CTX-M-1 enzymes are the common ESBLs in E. coli isolated from both sick and healthy broilers in several European countries (EFSA, 2011).

In Norway

Norway is one of the few European countries with a low third-generation cephalosporin resistance rate in E. coli from broilers. In general, the rate of third-generation cephalosporin resistant E. coli from broilers has ranged from 0-1.5% (ECDC & EFSA, 2016). However, this rate dropped to 0 % in 2016 (ECDC &EFSA, 2018). Norway achieves this low resistance level due to negligible use of antimicrobials in its broiler production pyramid and maintaining high biosecurity level at the broiler farms (NORM/NORM-VET, 2006, 2009, 2012).

In Europe

Unlike Norway, third-generation cephalosporin resistance in E. coli from broilers is higher in most European countries, with the exception of Sweden, Denmark, and Iceland (Mo, 2016).

In 2014, the overall resistance rate ranged from 0-32.2% with countries like Malta, Slovakia, and Spain showing moderate levels of resistance between 12.9-15% (Figure 7). High

resistance levels (>30%) were seen in countries including Cyprus, Latvia and Lithuania (EFSA and ECDC, 2016). Compared to 2014, data from 2016 showed that the majority of the countries had significant decrease in resistance to third-generation cephalosporin (cefotaxime) in E. coli from broilers (Figure 8). However, in Lithuania, the level increased from 31.8% in 2014 to 52% in 2016 (EFSA and ECDC, 2016; EFSA and ECDC, 2018).

Figure 7. Prevalence of cephalosporin resistance in indicator E. coli isolated from European broiler in 2014.

Adapted from (EFSA and ECDC, 2016).

Figure 8. Occurrence of cephalosporin-resistant E. coli isolates from broilers in European Union in 2016.

Adapted from (EFSA and ECDC, 2018).

1.9 Public health: ramifications of ESBL-producing E. coli

ESBL-producing E. coli can cause infections in humans such as bacteremia, sepsis, and urinary tract infections (Collignon et al., 2013). Owing to their resistance, these bacteria reduce the effectiveness of third-generation cephalosporins. Thus, this can lead to longer hospital stays and higher risks of nosocomial and community-acquired infections (Seiffert et al., 2013). Since ESBL-producing E. coli occurs in food and food producing animals, food as a possible transmission route for ESBL-producing bacteria to humans remains an open question. However, there has been some inconclusive evidence concerning the direct transmission of ESBL-producing bacteria from food-producing animals to humans, through the food chain (Collignon et al., 2013). This transmission is linked to handling or

consumption of contaminated food, which may serve as a reservoir for ESBL-producing bacteria (Coque et al., 2008).

Leverstein-van Hall et al., (2011) and Overdevest et al., (2011) discovered identical E.

coli strains from chicken meat and humans. However, results from both studies were based on molecular typing methods that have low discriminatory power. By contrast, de Been et al.,

(2014) and Berg et al., (2017) performed a whole genome sequencing analysis (WGS) to study E. coli strains from humans and chicken meat. This method generally has the highest level of resolution to discriminate between isolates and yields more reliable results. de Been et al., (2014) found no close relationship between the E. coli strains from chicken meat and humans as previously reported by Leverstein-van Hall et., (2011). However, identical

plasmids, i.e. IncI1α/ST3 and IncI1α/ST7 that disseminate the blaCTX-M-1 gene was discovered (de Been et al., 2014). On the other hand, WGS analysis applied in a Norwegian study

showed the E. coli strains from patients suffering from UTI infection and chicken meat were closely-related, including an IncK plasmid with blaCMY-2 (Berg et al., 2017). Hence, by considering the results from Berg et al., (2017), clonal transfer of cephalosporin-resistant E.

coli from chicken meat to humans can be possible to some extent.

1.10 Aim of study

The aim of this study was to investigate how ESBL-producing E. coli with blaCTX-M-1

emerged and disseminated in the Norwegian broiler production. Three hypotheses regarding the emergence and dissemination of these bacteria were investigated;

(1) a possible clonal spread of a specific E. coli variant with one blaCTX-M-1 harbouring plasmid (clonal spread)

(2) more E. coli clones with different blaCTX-M-1 harbouring plasmids, and

(3) horizontal transmission of a specific blaCTX-M-1-harboring plasmid to different E. coli strains from broilers in Norway.

2 Materials and methods.

All laboratory experiments were performed at the Norwegian Veterinary Institute (NVI) in Oslo. Bacterial isolates included in the study were previously from boot swabs collected in all broiler flocks sampled the Norwegian Salmonella control program from May-October 2016, and available for further characterization.

2.1 Selective isolation and confirmation of ESBL-producing E. coli

Boot swab samples were homogenized in peptone water and 100 µl inoculated on a MacConkey agar with 1 mg/mL cefotaxime to select cephalosporin resistant isolates. Plates were incubated overnight at 37 °C. Colonies with similar morphology and showing

cefotaxime resistance were isolated from the MacConkey agar, sub-cultured onto blood agar and incubated at 37 °C overnight. A matrix laser desorption ionization-time of flight

(MALDI-TOF) technique was used to confirm the isolates from the blood agar plates as E.

coli (MALDI-TOF, Bruker Daltonics).

The E. coli isolates were confirmed as ESBL producers based on the synergy test between third-generation cephalosporin and clavulanic acid. This was carried using the automated Sensititre^TM ESBL/AmpC MIC plates (TREK Diagnostic Systems), and results interpreted according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST) clinical breakpoints. In addition, PCR confirmed the presence of blaCTX-M-1 as the ESBL genotype in all the isolates. Isolation and identification of E. coli as ESBL producers was performed by the laboratory staff at NVI.

2.2 Characterization of ESBL-producing E. coli isolates

A total of 35 ESBL-producing E. coli with blaCTX-M-1 were included in this study. Of these, 26 were from broiler flocks and seven from parent flocks. In addition, two ESBL- producing E. coli isolates from poultry in Iceland were included. The two isolates from Iceland were confirmed as ESBL-producers in this study. In addition to typing of the isolates, their susceptibility to different antimicrobial agents was determined.

2.2.1 DNA extraction

DNA was extracted from all 35 isolates by the boiling lysis method. Bacterial suspensions were prepared in 100 µl milli-Q water using a single colony from blood agar. Cells were lysed at 100 °C for 15 minutes and the suspension centrifuged for 5 minutes at 13200 rpm. The supernatants were transferred to new Eppendorf tubes and used as DNA template.

2.2.2 Phylogenetic grouping

A multiplex PCR method (Doumith, Day et al. 2012) was used to assign phylogenetic groups to the 35 E. coli isolates. Primers used to amplify conserved regions of the four

phylogenetic markers: gadA, chuA, yjaA, and DNA fragment TSPE4.C2 are shown in Table 1.

A phylogroup B2 E. coli strain was used as the positive control as it contains all four phylogenetic markers (Sunde et al., 2015). Table 2 illustrates the assignment of each isolate into the respective phylogenetic group based on the combinations of the four markers.

Table 1. Primer sequences used for the amplification of target regions and phylotyping of E.

coli isolates.

Genetic marker Primer sequence (5ʹ 3ʹ) Base pair (bp) length

gadA Forward:

GATGAAATGGCGTTGGCGCAAG Reverse:

GGCGGAAGTCCCAGACGATATCC

373

chuA Forward:

ATGATCATCGCGGCGTGCTG Reverse:

AAACGCGCTCGCGCCTAAT

281

yjaA Forward:

TGTTCGCGATCTTGAAAGCAAACGT Reverse:

ACCTGTGACAAACCGCCCTCA

216

TSPE4.C2 Forward:

GCGGGTGAGACAGAAACGCG Reverse:

TTGTCGTGAGTTGCGAACCCG

152

In an Eppendorf tube, 10 µL of each primer (10 µM) for the four markers were mixed

together with 20 µL Milli-Q water to obtain a “primer mix”. The PCR reaction of each isolate was carried out in a 25 µL reaction mixture containing 12.5 µL 1x Qiagen Multiplex PCR mix, 0.5 µL 0.2 µM primer mix, and 2 µL of DNA template. Next PCR was run using Sure cycler 8800 (Agilent Technologies) under the conditions shown in Table 3. Ten microliters of each PCR amplification product was visualized with 2.5 µL 6x DNA loading dye and

separated on a 1% agarose gel stained with 10 µl GelRed nucleic acid dye (ThermoFisher Scientific).

Table 2. Interpretation of the phylogenetic grouping of the E. coli isolates based on the presence and/or absence of the four phylogenetic markers.

Phylogenetic group

gadA chuA yjaA TSPE4.C2

A + - +/- -

B1 + - - +

B2 + + + +/-

D + + - +/-

Table 3. Multiplex PCR program Hold for 4 minutes at 95 °C 30 cycles 30 seconds at 95 °C 30 seconds at 60 °C 30 seconds at 72 °C Hold for 5 minutes at 72 °C Infinity 8 °C

2.3 Antimicrobial susceptibility testing

Antimicrobial susceptibility testing was performed by the broth microdilution method and agar diffusion method.

2.3.1 Micro broth dilution for MIC determination

Minimum inhibitory concentration (MIC) values were determined for the 14 antimicrobial agents sulfamethoxazole (SMX), trimethoprim (TMP), ciprofloxacin (CIP), tetracycline (TET), meropenem (MERO), azithromycin (AZI), nalidixin acid (NAL), cefotaxime (FOT), chloramphenicol (CHL), tigecycline (TGE), ceftazidime (TAZ), colistin (COL), ampicillin (AMP), and gentamicin (GEN). The procedure was carried out using the Sensititre™ Gram Negative MIC Plate (ThermoFisher Scientific).

MIC values (mg/l) of the antimicrobials had been determined against 33 of the 35 E. coli isolates (personal communication, Solveig Sølverød Mo, NVI). Thus, in this study, MIC values of the 14 agents were determined against the two E. coli isolates from Iceland.

In brief, bacterial suspensions with a 0.3-0.5 McFarland were prepared separately for the isolates in a 5 mL distilled water. McFarland. Fifty-microliters of the suspension was

inoculated into 11 mL Sensititre ® Cation Adjusted Mueller-Hinton Broth (CAMHBT) and 50 µl of the mixture automatically inoculated into each well in the microtiter plates using the

Sensititre® AIM^TM pipetting robot. Wells in the microtiter plates are dosed with different concentrations of the above mentioned antimicrobial agents. Plates were incubated at 35 ± 1°C °C overnight. E. coli ATC25922 isolate was used as the quality control strain.

2.3.2 Agar disk diffusion method

Bacterial suspensions were made in a 5 mL 0.9% saline water and turbidity adjusted to 0.3-0.5 McFarland. A cotton swab dipped in the suspension was spread on a Mueller-Hinton agar plate (Oxoid^TM, ThermoScientific^TM) using an automatic plate rotator. Antimicrobial discs were placed onto the plates using the Oxoid^TM Antimicrobial Susceptibility Disc Dispenser (ThermoFisher Scientific) and incubated at 35 ± 1°C overnight. The antimicrobial discs included: ampicillin (10 µg), amoxicillin/clavulanic acid (30 µg),

sulfamethoxazole/trimethoprim (23.75 +1.25 µg), tetracycline (30 µg), cephalexin (30 µg), ciprofloxacin (5 µg), neomycin (30 µg), gentamicin (10 µg), polymixin/colistin (300 units), trimethoprim (5 µg), nalidixic acid (30 µg), clindamycin (2 µg), erythromycin (15 µg), and penicillin (1 unit). E. coli ATC25922 was used as a quality control strain. Inhibition zone diameters (mm) were interpreted according to the EUCAST Clinical breakpoints (version 7.1;

2017-03-10).

2.4 Pulsed-Field Gel Electrophoresis (PFGE)

The genetic relatedness between the 35 E. coli isolates was determined using the pulsed- field gel electrophoresis protocol described in (Caprioli et al., 2014). The procedure involves:

1) preparation of bacterial cell suspension 2) preparation of agarose plugs 3) plug lysis 4) plug washing 5) restriction enzyme digestion 6) gel electrophoresis 7) staining and visualization of the gel, and 8) data analysis.

2.4.1 Preparation of cell suspension

Cell suspensions was prepared for each of the 35 E. coli isoltes in 2 mL Tris-EDTA (TE) buffer and turbidity measured to an optical density (OD) of 0.7-0.79.

2.4.2 Gel plug preparation

Of each cell suspension, 400 µL was mixed with 20 µL proteinase K and 400 µl 1%

melted PFGE agarose gel. The mixture was transferred into a PFGE disposable plug mold (Bio-Rad Laboratories) to generate agarose plugs.

2.4.3 Lysis and washing of plugs

Bacterial DNA embedded in the agarose plugs was lysed in a mixture of 25 µl Proteinase K and 5 mL cell lysis buffer (CLB 0.1 mg/mL) at 55 ^°C for two hours. After lysis, plugs were washed twice in 10 mL pre-heated milliQ water and four times in 10 mL pre-heated TE- buffer. Both milliQ water and TE-buffer were pre-heated at 50 ^°C. Between each wash, mixture was placed in a shaking incubator for 15 minutes at 50 ^°C. Gel plugs were transferred into 1 mL cold TE-buffer and stored at 4 ^°C.

2.4.4 Digestion of DNA in agarose plugs

Gel plugs were cut into 2-2.5 mm slices with a glass coverslip and each slice digested with 5 µL 10U/µL XbaI restriction enzyme (ThermoFisher Scientific) for 3.5 hours at 37 ^°C.

2.4.5 Gel electrophoresis

Digested plugs slices were loaded into wells of a 14 x 14 cm gel form and separated in a CHEF DR III system (Bio-Rad Laboratoratories, Hercules, CA) on a 1% SeaKem Gold PFGE grade agarose. Electrophoresis ran for 24 hours under the following conditions described in (Agersø et al., 2014). An XbaI-digested DNA from Salmonella enterica serovar Braenderup strain H9812 was used as the marker.

2.4.6 Staining and visualization of the gel

The gel was stained in a mixture of 120 µl GelRed + 400 mL Milli-Q water for 25 minutes and de-stained in 400 mL Milli-Q water for 15 minutes. A Bio-Rad Molecular Imager® Gel Doc^TM XR+ imaging system (Bio-Rad Laboratories, Milan, Italy) was used to visualize the gel.

2.4.7 Data analysis of the gel image

The BioNumerics software v 6.6 (Applied Maths, …)was used to analyze the generated PFGE fingerprints. The similarities of fingerprints were compared using a Dice correlation coefficient at 1.5% tolerance and 1.5% optimization, and a dendogram constructed with the unweighted pair group method with arithmetic averages (UPGMA) clustering method using the BioNumerics program (Agersø et al., 2014). Isolates with similarity at ≥ 97% cut-off value were considered as clonally-related whereas isolates with similarity of ≥ 80% were grouped in the same PFGE cluster (Mo et al., 2016).

2.5 Conjugative transfer and characterization of plasmid replicons

2.5.1 Conjugation experiments

Transfer of blaCTX-M-1-carrying plasmids was determined in a series of conjugation broth mating experiments as described in (Mo et al., 2016). The cephalosporin-resistant E. coli isolates, which were sensitive to nalidixic acid (naI^S) served as the donor strains whereas a plasmid-free E. coli DH5α (nalidixic acid resistant) was used as the recipient strain.

Overnight cultures of both donor and recipient strains were prepared separately in 4 mL Luria Bertani (LB) broth at 37°C. Next, 500 µL of recipient strain was mixed with 10 µL of each donor strain in a 4 mL LB broth and incubated at 37°C without shaking for four hours.

After four hours, 100 µL of broth mating was plated on a Mueller-Hinton (MH) agar supplemented with 0.5 mg/L cefotaxime and 20 mg/L nalidixic acid to select the transconjugants. This step was repeated after 6 and 24 hour-broth mating if no transconjugants were obtained after the 4 hours.

2.5.2 PCR-detection of blaCTX-M-1 in the transconjugants

To examine if blaCTX-M-1 transferred from the donor strains to the recipient strain, PCR was used to confirm the presence of the blaCTX-M-1 in the transconjugants. The blaCTX-M-1 gene was targeted using the primers (forward: 5ʹ ATGTGCAGYACCAGTAARGTKATGGC 3ʹ and reverse: 5ʹ TGGGTRAARTARGTSACCAGAAYCAGCGG 3ʹ) described in (Hasman et al., 2005).

Genomic DNA was extracted from transconjugants using the boiling lysis method.

Extracted DNA was used as the template for the PCR reaction, which was carried out in a 25 µL mixture containing 2.5 µL 1x PCR buffer, 0.5 µL0.2 mM dNTP mix, 0.5 µL 0.2 µM of each blaCTX-M-1 primer, 0.1 µL 0.5 U Taq DNA polymerase, 18.4 µL milli-Q water, and 2.5 µL of extracted DNA. Table 4 shows the amplification conditions for the PCR reactions (Agilent Surecycler 8800). To visualize the presence of the blaCTX-M-1, PCR products and 6x loading dye (LD) were mixed in a 10 µL: 2.5 µL ratio and run on a 1% (w/v) agarose gel electrophoresis. E. coli K8-1 strain and MilliQ-water was used as positive and negative control, respectively.

Table 4. Thermal cycler conditions for PCR detection of blaCTX-M-1. Hold for 5 minutes at 95 °C

30 cycles 30 seconds at 95 °C 30 seconds at 60 °C 1 minute at 72 °C Hold for 7 minutes at 72 °C Infinity 8 °C

2.5.3 Plasmid typing: PCR-based replicon typing (PBRT) of transconjugants Transconjugants positive for blaCTX-M-1 were subjected to plasmid replicon typing using the commercial PCR-based replicon typing (PBRT) kit (Diahtheva, Italy). PBRT determines the Inc groups of major plasmid families in Enterobacteriaceae (Liebana et al., 2013). In principle, the PBRT kit is composed of eight specific standard PCR assays (M1-M8) (Figure 9). Primers in one PCR assay can target and amplify three to four amplicons that represent major plasmid Inc groups located on resistance plasmids among Enterobacteriaceae (Carattoli et al., 2005). In addition, the kit contains positive controls for each PCR mix.

Figure 9. PCR organization and replication targets in the PBRT-KIT scheme. Each PCR mix is color-coded.

Adapted from PBRT-kit (version 14/02/2017) (Diatheva, Fano, Italy).

The PBRT procedure was performed according to the manufacturer’s instructions

(Diatheva, Fano, Italy). First, a mastermix solution consisting of each PCR mix (i.e. M1) and DNA polymerase (5U/µL) was prepared and 24 µL of the mastermix aliquoted into PCR strip tubes. Next, 1 µL of positive control was added to one PCR strip tube and 1 µL of each transconjugant DNA template to the remaining tubes. The PCR reaction was run under the conditions shown in Table 5. All amplification products were visualized by gel

electrophoresis on a 2.5% agarose gel stained with 20 µl GelRed. One microliter milliQ water was used as the negative control.

According to the PBRT manufacturer’s protocol, IncK plasmids can react with both B/O and K primers in the M2 and M7 PCR mixes. To discriminate between Inc B/O and K plasmids, a PCR that specifically targets the IncB/O replicon was performed on isolates showing weak Inc K amplicon using the PBRT kit. The Inc B/O primers and PCR conditions used is described previously (Carattoli et al., 2005).

Table 5. Thermal cycler condition of the PBRT-kit

1 cycle 95℃ for 10 min

25 cycles 95℃ for 60 secs

60℃ for 30 secs 72℃ for 60 secs

1 cycle 72℃ for 5 min

Cool down to 4℃

2.5.4 Suceptibility testing of transconjugants

Following conjugation was the phenotypic testing of the transconjugants to six antimicrobial agents by disk diffusion described in 2.3.2. This was done to detect the co- transfer of resistance genes other than the blaCTX-M-1. The antimicrobial discs used included ampicillin (10 µg), tetracycline (30 µg), trimethoprim (5 µg), sulfamethoxazole, ceftazidime, and cefotaxime.

2.6 Whole genome sequencing (WGS)

All 35 isolates had previously been subjected to whole genome sequencing, and assembled sequences were available for this study. Following is a brief description of the methods used.

2.6.1 Bacterial DNA isolation with Qiasymphony

Genomic DNA isolation with QiAsymphony DSP DNA mini kit and QIAsymphony SP automated instrument (Qiagen® Sample &Assay Technologies) was performed by Solveig Sølverød Mo. The concentrations (ng/µL) of the extracted DNA were determined with a Qubit® dsDNA BR Assay Kit (ThermoFisher Scientific), and purity measured with a NanoDrop^TM 2000 UV spectrophotometer (ThermoFisher Scientific).

2.6.2 Sequencing

Genomic DNA was sequenced on a NextSeq 500 Illumina platform, obtaining 150 bp paired-end reads. Sequencing of the isolates was performed at the Norwegian Sequencing Centre, Ullevål, Oslo.

Before sequencing, a Nextera XT DNA Library preparation kit (Illumina, USA) was used to prepare the sample libraries. In principle, the Nextera XT DNA Library preparation

workflow occurs in six steps and involve: (1) tagging of genomic DNA, (2) cleaning the tagged DNA (3) amplification of libraries (i.e. tagged DNA fragments), (4) cleaning up the libraries, (5) checking the libraries, and (6) normalization and pooling of the libraries.

First, the extracted genomic DNA normalized to 0.2 ng/µl, are fragmented and tagged with adaptor sequences. Tagged DNA is purified through Zymo DNA binding buffer and Zymo DNA Wash Buffer. Purified tagged DNA is amplified using a 5-cycle PCR program and the DNA libraries produced are purified with an AMPure XB beads. Purified libraries are quality-controlled on an Agilent Technology 2100 Bioanalyzer. In the last step, the libraries are normalized to 2 nM and pooled, i.e. mixed together in a single tube, which is then diluted and de-natured before sequencing. See the Illumina Nextera® DNA Library Prep Reference Guide (1000000006836 v00, January 2016) for more information about library preparation protocol.

2.6.3 Pre-processing of raw sequence data: assembly of genomes

Raw reads generated after sequencing were pre-processed to yield high quality data for analysis. Reads were quality controlled using the FastQC tool and trimmed to remove duplicate reads and adaptor sequences using Trimmomatic (Bolger et al., 2014). Reads were de novo assembled into contigs using SPAdes version 3.9.0 (Bankevich et al., 2012) and assemblies evaluated with the QUAST assembly tool (Gurevich et al., 2013). The complete genome of the E. coli K-12 substr. MG1655 strain (Accession number NZ_CP027060.1) was used as a reference. Pre-processing of the raw sequence reads was performed by Camilla Sekse, a researcher at NVI.

2.7 In silico analysis of the whole genome sequencing:

2.7.1 Bacterial Analysis Pipeline:

The 35-assembled sequence reads in FASTA files were uploaded to the web-based Bacterial Analysis Pipeline (BAP) for data analysis. The BAP is an automatic and robust tool

that analyses bacterial genomes (Thomsen et al., 2016). This pipeline, with its default settings, executes a series of workflow involved in the analysis of bacterial isolates (Figure 10).

During the workflow, if uploaded sequences are unassembled, the BAP begins with a draft de novo assembly of the sequence reads into contigs (Thomsen et al., 2016). The KmerFinder tool, which is run in parallel to the Assembly, identifies species. The ContigAnalyzer tool, analyzes assembled contigs by calculating the number of contigs, total number of bases, and the N50 value, which is defined as the shortest contig (sequence length) that represent 50% of the whole genome assembly (Larsen et al., 2012). After the ContigAnalyzer service, the assembled contigs are submitted to ResFinder for identification of acquired resistance genes (Joensen et al., 2014). BAP gradually performs the remaining services, which includes multi- locus sequence types (MLST) (Larsen et al., 2012) PlasmidFinder and pMLST (Carattoli et al., 2014, Thomsen et al., 2016), and VirulenceFinder (Joensen et al., 2014). All these services, available at the Center for Genomic Epidemiology, DTU, Denmark:

https://cge.cbs.dtu.dk/services/ are run based on their databases that relate to E. coli.

Figure 10. Bacterial Analysis Pipeline workflow. Adapted from (Thomsen et al., 2016).

2.7.2 Phylogenetic analyses: CSI-phylogeny

Phylogenetic analysis of the isolates was based on the single nucleotide

polymorphisms (SNPs) difference between the whole genome sequenced data. SNPs were determined using the CSI Phylogeny v 1.4 web tool available at

https://cge.cbs.dtu.dk/services/CSIPhylogeny/. The assembled WGS data were uploaded and analysis run with the default parameters of the pipeline as described in (Kaas et al., 2014).

The reads from each genome was mapped against the reference genome: E. coli str. K.12 substr. MG1655 (Accession number NZ_CP027060.1). A Newick file generated after the SNP analysis was used to construct a phylogenetic tree on Figtree v1.4.3.

2.7.3 Serotyping (SerotypeFinder)

Serotypes of the sequenced E. coli isolates were identified with the SerotypeFinder v 1.1 web tool: https://cge.cbs.dtu.dk/services/SerotypeFinder/ (Joensen et al., 2014)

2.7.4 blaCTX-M-1-harboringplasmid characterization

In addition to plasmid typing and subtyping, one blaCTX-M-1-carrying plasmid was reconstructed to determine its nucleotide sequence. As several studies have reported the prevalence of highly similar IncI1α plasmids carring blaCTX-M-1 in the broiler production pyramid (Touzain et al., 2018, Wang et al., 2014), it was hypothesized that the blaCTX-M-1- carrying plasmids from this study share similarities to plasmids found in broilers from other European countries.

To characterize the plasmid, the contig sequence that contained the blaCTX-M-1- harboring plasmid was extracted from the WGS data using CLC Genomics (CLC Bio, Qiagen, Aarhus, Denmark). The contig containing the blaCTX-M-1-harboring plasmid was aligned with two plasmids from Switzerland (accession no. KM377238 and KM377239) and the comparisons visualized using BLAST Ring Image Generator (BRIG) (Alikhan et al., 2011). Another plasmid (accession: SAMN07197432) from France was also aligned with the contig containing the blaCTX-M-1-harboring plasmid. Primers were designed in CLC Genomics to determine sequences of gaps pointing outward from the ends of the contigs.

Using the primers, a gradient PCR was performed to amplify the target sequences followed by Sanger sequencing of the amplified PCR products. In general, gradient PCR is a series of individual PCR reactions of the same content (i.e. DNA, primers, enzymes and buffers) performed at different annealing temperature ranges. The idea here was to generate

overlap sequences that can close the gaps and obtain a complete sequence of the contig with the blaCTX-M-1-harboring plasmid.

2.7.4 i DNA extraction with QiaAMP DNA kit

DNA was extracted from donor and the corresponding transconjugants of the 2016-40- 14272 isolate using QiaAMP DNA mini kit according to the manufacture’s protocol

(Qiagen®). DNA concentrations was quantified using Qubit® dsDNA BR Assay Kit (ThermoFisher Scientific), and purity measured with a NanoDrop^TM 2000 UV

spectrophotometer (ThermoFisher Scientific). The extracted DNA were used as template for the PCR reaction.

2.7.4 ii PCR and Sanger sequencing

The PCR reaction was carried out in a 25 µL mixture containing 2.5 µL Qiagen 10x buffer, 0.5 µl dNTP mix, 1 µL 0.4 µM forward primer, 1 µL 0.4 µM reverse primer, 0.1 µL Qiagen Taq polymerase, 18.9 µL milliQ water, and 1 µL of template DNA. Table 6 shows the amplification conditions for the gradient PCR reaction run on the T100^TM thermal cycler (BIO-RAD). To visualize the products formed, 10 µl of each PCR product was mixed with 2.5 µL 6x loading dye (LD) and run on a 1% (w/v) agarose gel electrophoresis. Two PCR products with high gel band intensity were sent for Sanger sequencing. Sanger sequencing of the PCR products was performed by the staff at the molecular biology section at NVI.

Table 6. Program for the gradient PCR set-up Hold for 5 minutes at 95 °C

25 cycles 30 seconds at 95 °C *30 seconds at 50-60 °C 1 minute at 72 °C Hold for 2:30 minutes at 72 °C Infinity 8 °C

* Annealing temperature ranges: 50-60 °C

2.7.5 Annotation of the plasmid

The nucleotide sequence of the contig with blaCTX-M-1-plasmid was annotated

automatically using the Online Rapid Annotation Subsequencing Technology (RAST; Aziz et al., 2008) and manually in CLC Main Workbench visualized in CLC genomics.

3. Results

3.1 Genotypic and phenotypic characterization of the E. coli isolates 3.1.1 Phylogenetic grouping

The 35 E. coli isolates were assigned to three phylogenetic groups, i.e. A, B1, and D.

Of these, 17 isolates (48%) belonged to the virulent extra-intestinal E. coli group D. Ten (29%) and eight (23%) isolates were classified into the commensal phylogenetic groups A and B1, respectively (Figure 11 B). Figure 12 depicts the assignment of nine selected isolates to their respective phylogenetic group.

In broilers, the phylogroup D was most represented among the E. coli isolates. Of the 26 isolates, 15 (58%) were classified as phylogroup D whereas eight (31%) and three (11%) isolates belonged to phylogroup B1 and A, respectively. All seven E. coli isolates from parent flocks, by contrast, belonged to phylogroup A (Figure 12 A and C). The two poultry samples from Iceland with unknown origin were assigned to the phylogenetic group D and included in Figure 12 B.

Figure 11. Distribution of ESBL-producing E. coli isolates from broilers and parent. All 35 isolates divided into the phylogenetic groups A (n= 10) , D (n= 17), and B1 (n= 8) is shown in the middle (B).

Figure 12. Multiplex PCR profile demonstrating phylogrouping of E. coli isolates from nine of the poultry samples. Lanes 1, 2, 7, and 9: phylogroup A [ gad+, chuA-, yjaA+, TSPE4.C2-]; lanes 3 and 5: phylogroup D [gad+, chuA+, yjaA, TSPE4.C2+]; lanes 4, 6, and 8: phylogroup B1 [gadA+, chuA-, yjaA-, TSPE4.C2+]. Lanes 10 and 12: phylogroup B2-E. coli as positive control [gadA+, chuA+, yjaA+, TSPE4.C2+]. Lane 11: negative control and lanes M: GeneRuler^TM 50 bp DNA ladder.

3.1.2 Antimicrobial resistance profile: phenotypic testing

Based on the MIC and disk diffusion tests, some isolates displayed multi-drug resistance (MDR) phenotype. MDR symbolizes bacteria that resist three or more different classes of antimicrobials (DANMAP, 2016).

3.1.2.i MIC determination

In the susceptibility testing of the isolates to the 14 antimicrobial agents, occurrence of resistance was highest to ampicillin cefotaxime, ceftazidime, and sulfamethoxazole. Table 7 shows distribution of the MIC values and level of antimicrobial resistance among the 35 ESBL-producing E. coli isolates. Overall, resistance to ampicillin (MIC= >64 mg/L) and cefotaxime (MIC= >4 mg/L) was 100% whereas ceftazidime (MIC= 1-2 mg/L) and sulfamethoxazole (MIC= >1024 mg/L) resistance was 97.1% and 91.4%, respectively. Six isolates (17%) were resistant to tetracycline (MIC= ≥64%) and three isolates (8.6%) to trimethoprim (MIC= >36 mg/L). None of the isolates showed resistance phenotype to the remaining antimicrobials: ciprofloxacin (MIC= ≤0.015-0.03 mg/L), nalidixic acid (MIC=

≤4mg/L), tigecycline (MIC= ≤0.25 mg/L), colistin (MIC= ≤1 mg/L), gentamicin (MIC= ≤0.5- 1 mg/L), and the carbapenem drug- meropenem (MIC= ≤0.03 mg/L).